Colon carcinoma cells show reduced MARCKS expression or enhanced MARCKS phosphorylation
The present investigation is based on the hypothesis that MARCKS—via its ability to bind phosphatidylinositol 4,5-bisphosphate (PI(4,5)P2)—affects the function of the ABCB1 transporter.
In this context, it is noteworthy that the PI(4,5)P2-binding function of MARCKS is regulated by its phosphorylation state: Whereas non-phosphorylated MARCKS is associated with the plasma membrane and interacts with PI(4,5)P2, phosphorylated MARCKS (phospho-MARCKS) is translocated to the cytoplasm and does not interfere with PI(4,5)P2 (Fig. 1a). Thus, with respect to its role in PI(4,5)P2-binding (and irrespective of other, PI(4,5)P2-independent MARCKS functions) phospho-MARCKS can be designated as inactive, whereas non-phosphorylated MARCKS acts as the active variant (Fig. 1a).
Of note, lack of MARCKS expression in colorectal cancer (CRC) has been associated with a more aggressive tumor phenotype and unfavorable prognosis (Chen et al. 2014, 2015; Rombouts et al. 2013), suggesting a tumor suppressor function of MARCKS (Rombouts et al. 2013). Thus, if the antitumor effects of MARCKS are dependent on its ability to interact with PI(4,5)P2, CRC cells with hyperphosphorylated (inactive) MARCKS should behave like CRC cells with absent MARCKS. Therefore, we performed immunohistochemistry analyses of CRC tumor samples from nine patients to establish the presence of phosphor-MARCKS in CRC. In fact, MARCKS staining (total MARCKS) was absent in three tumor samples and markedly reduced in another three samples in contrast to strong MARCKS staining in adjacent normal tissue (Fig. 1b, left). This finding is in line with the aforementioned postulate of a tumor-suppressive role of MARCKS in CRC. However, in three further samples increased phospho-MARCKS signals were detectable as compared to the adjacent normal tissue (Fig. 1b, right). This provided the rationale to elucidate the potential role of phosphorylated MARCKS in CRC.
For this purpose, we used in the following investigations the two well-established human CRC model cell lines LoVo and HT-29 (Fig. 1c). These two cell lines are particularly suitable for the functional characterization of MARCKS in the CRC context, since LoVo cells completely lack MARCKS expression owing to a genomic deletion (Bickeböller et al. 2015; Rombouts et al. 2013) and HT-29 cells show heavily phosphorylated MARCKS (Bickeböller et al. 2015) (Fig. 2c, d).
Evidence for a connection between MARCKS expression and ABCB1 function using MARCKS-negative LoVo cells
To further test whether MARCKS negatively affects ABCB1 we used MARCKS-negative LoVo cells and introduced wild-type MARCKS with a flag tag (MARCKS-WT-flag) or effector domain-negative flag-tagged MARCKS (MARCKS-ΔED-flag), which does not bind PI(4,5)P2 (Fig. 2a).
Expressions of wild-type MARCKS, but not of inactive, effector domain-mutated MARCKS increased the uptake of the ABCB1 substrate calcein into LoVo cells, indicating a reduced function of the ABCB1 efflux pump (Fig. 2b). This effect of MARCKS expression on calcein uptake was comparable with the impact of the well-established ABCB1 blocker verapamil (Fig. 2b), indicating the potential relevance of this mechanism.
Moreover, MARCKS and ABCB1 expression were inversely correlated in six CRC tumor samples with reduced or absent MARCKS expression and in adjacent normal tissue: Normal tissue showed high MARCKS and low ABCB1 expression and tumor tissue displayed low MARCKS and high ABCB1 expression (Fig. 2c). Of note, this inverse correlation of MARCKS and ABCB1 expression was also observed in three tumor samples with highly phosphorylated MARCKS (data not shown), suggesting that the phosphorylation status of MARCKS is relevant for its impact on ABCB1 function.
Based on these findings, we next made use of a MARCKS mutant with absent phosphorylation sites (MARCKS-S4A) and a MARCKS mutant with phospho-mimicking mutations (MARCKS-S4D). In cells expressing MARCKS-S4A, the inhibition of MARCKS function via phosphorylation (cf. Fig. 1a) is abrogated. In contrast, the MARCKS-S4D mutant behaves like inactive, hyperphosphorylated MARCKS. In line with these assumptions, LoVo cells expressing wild-type MARCKS showed a mixed subcellular distribution of MARCKS both in the cytoplasm and in association with the plasma membrane (Fig. 3a, top panel). In contrast, the constitutively active MARCKS-S4A mutant was only expressed at the plasma membrane (Fig. 3a, middle panel) and the inactive (phospho-mimicking) MARCKS-S4D mutant was translocated into the cytoplasm (Fig. 3a, lower panel).
Using these MARCKS constructs, we again tested for the effect of calcein uptake in LoVo cells (Fig. 3b). In line with the findings with flag-tagged MARCKS (cf. Fig. 2b), both wild-type and constitutively active MARCKS (MARCKS-S4A) increased calcein uptake (suggestive of an ABCB1 inhibition), whereas the inactive MARCKS-S4D mutant had no effect on calcein accumulation (Fig. 3b).
Role of MARCKS phosphorylation for regulation of ABCB1 function
In HT-29 cells with hyperphosphorylated MARCKS showing a functional inactivation of MARCKS (cf. Fig. 1a) this inhibition represents a potential therapeutic option to modulate ABCB1 activity. First we tested for the expression of ABCB1 in HT-29 and LoVo cells via immunostaining and Western blot (Fig. 3c and d).
As previously shown (Jin et al. 2012; Kalwa et al. 2012, 2014; Kalwa and Michel 2011), MARCKS phosphorylation in endothelial cells is reduced by the tyrosine kinase inhibitor bosutinib. Bosutinib also inhibited MARCKS phosphorylation in HT-29 cells (Fig. 3d). As expected, LoVo cells showed no MARCKS expression (Fig. 3d). In the functional calcein assay both in HT-29 and LoVo cells the accumulation of calcein over time was accelerated upon addition of the well-established ABCB1 inhibitors tariquidar (0.1 µM) and verapamil (50 µM) (Fig. 3e) and both cell lines have detectable levels of ABCB1 expression. Notably, bosutinib (5 µM) treatment increased calcein accumulation in HT-29 cells in a similar manner as the direct ABCB1 blocker verapamil (3.2-fold increase in calcein uptake rate, p = 0.002; n = 35; Fig. 3e). In contrast, in LoVo cells, which are devoid of MARCKS (Fig. 3d), bosutinib did not exert any effect on calcein uptake, despite intact ABCB1 function as indicated by the verapamil and tariquidar positive controls (Fig. 3e). This indicates a direct role of MARCKS in ABCB1 regulation.
To assess the role of PI(4,5)P2, we used a FRET biosensor (Aoki et al. 2013) for PI(4,5)P2. Treatment of HT-29 cells with bosutinib markedly decreased the level of free PI(4,5)P2 in the plasma membrane, whereas bosutinib had no effect in MARCKS-negative LoVo cells (Fig. 3f, g). Of note, bosutinib treatment only affected the abundance of the MARCKS binding partner PI(4,5)P2, but not the level of PI(3,4)P2, which is not sequestered by MARCKS. To analyze the specificity of bosutinib treatment further towards protein kinases that are known to modulate MARCKS, we utilized different FRET bioprobes. As expected, bosutinib led to a strong reduction in c-Abl activity. In contrast, activity of PKC, an alternative modulator of MARCKS, was not affected by bosutinib treatment (data not shown).
Next, we evaluated the association between MARCKS function and ABCB1 activity by analyzing the intracellular localization in HT-29 cells. For this purpose, we generated stable cell lines expressing low levels of MARCKS-eGFP or ABCB1-eGFP fusion proteins. Confocal microscopy revealed that MARCKS-eGFP was predominantly localized in the cytosol, whereas ABCB1 was found at the membrane. Notably, upon treatment with bosutinib MARCKS relocalized to the plasma membrane (Fig. 4a). In parallel ABCB1 was translocated to intracellular compartments (Fig. 4b).
It is known that the clathrin dependent protein internalization requires the precise removal of PI(4,5)P2 from the protein dynamin at the endocytotic region. We wanted to test if ABCB1 inhibitory effects of bosutinib are based on accelerated endocytosis. We employed pharmacological inhibitors against several key players of endocytosis in our calcein assay system. The bosutinib-induced reduction of ABCB1 function was reversed by chlorpromazine (50 µM), a poorly specific inhibitor of clathrin-mediated endocytosis (n = 31; p = 0.02), or by dynasore (80 µM), a direct inhibitor of dynamin-induced vesicle separation (n = 31; p = 0.03); (Fig. 4c). We wanted to see if this mechanism resulted in a prolonged reduction of ABCB1 abundance. Here (Fig. 5a), Western-blot experiments revealed that bosutinib treatment led to the significant reduction of ABCB1 expression in HT-29 cells.
Chemosensitiziation of HT-29 cells upon inhibition of MARCKS phosphorylation
In the light of these findings and given the fact that CRC cell sensitivities towards cytostatics can be dependent on ABCB1, a long-term bosutinib treatment may sensitize HT-29 cells to commonly used chemotherapeutics.
To explore this possibility, we performed cell proliferation assays using the WST-1 tetrazolium dye. We chose doxorubicin and 5-FU as chemotherapeutic agents, because there are either the classical example for ABCB1 function (doxorubicin) or broadly used in the treatment of CRC (5-FU). Both drugs are known for being transported via ABCB1 (Borst and Schinkel 2013; Chufan et al. 2015; Crowley et al. 2010). Upon single treatment of cells with bosutinib, doxorubicin or 5-FU, profound anti-proliferative effects were observed (Fig. 5b). Importantly, the combination of both agents with bosutinib displayed even greater effects, essentially abolishing cell proliferation. This was seen at an even greater magnitude in colony forming assays. Here upon a combination treatment with bosutinib and either doxorubicin or 5-FU reduced the colony numbers by 83% ± 9% (doxorubicin) or 72% ± 12%; (5-FU), respectively (n = 6; Fig. 5c–f). Taken together, this demonstrates that sensitivity to chemotherapy treatment is markedly increased or restored by addition of bosutinib, indicating that bosutinib treatment, through its effects in cases of hyperphosphorylated MARCKS, might be a promising therapeutic approach in preventing or overcoming chemotherapy resistance in CRC treatment.